Wind turbines, as every technological system, requires maintenance to ensure its sustainability and profitable life. But many accidents have been occurred since the beginning of installation of the first wind turbines.
According to Caithness Windfarms Information Forum (CWIF), an average of 33 accidents per year were observed between 1998 and 2002, 81 accidents per year between 2003 and 2007, 144 accidents per year between 2008 and 2012, and 167 accidents per year between 2013 and 2017 (http://www.caithnesswindfarms.co.uk/fullaccidents.pdf).
Maintenance is prone to errors as maintenance tasks are often complex, with limited information about the condition of every component and the tasks are performed as fast as possible. A non-reporting culture is obvious at many industry sectors as approximately 60% of the maintenance personnel surveyed by Hobbs and Williamson (2008) (https://www.tandfonline.com/doi/abs/10.1080/00140130210148528), reported that they had not mentioned a corrected error at company’s log files.
It is reported also in the US that many times wind turbines may be behind the required maintenance schedule mostly due to a shortage of qualified turbine technicians. In the wind turbine industry, failures are frequently justified as effects of overloading, overheating, weak designs, etc. but according to statistics from a manufacturing company, it is shown that 40% of wind turbine failures are due to human factors (http://archive.northsearegion.eu/files/repository/20120320111424_PC_Skills-Compendiuminmaintenance.pdf).
It is crystal clear that human plays an important role during the design, installation, production, maintenance, and operation phases of these systems. Human is responsible mainly for accidents and component failures. Human errors in maintenance can be detected as incorrect inspection, diagnosis, repair or installation of the equipment. For instance, incorrect calibration of equipment and wrong application of grease in critical components cause the most problems during maintenance according to experience.
The human factor has been tried to be analyzed by many scientists. Thus, many models had been developed such as the SHEL model, the error chain model, The Swiss cheese model by Prof. James Reason etc.
Generally, in literature there are 2 ways of thinking that tries to analyze human errors. The first one is called “Cognitive psychological school” represented by Prof. James Reason dealing errors as social act of life and the other one is called “Joint cognitive systems school” represented by J. Rasmussen, E. Hollnagel and D. Woods stating that the human error is not the cause but the symptom of other deeper problems inside the system and the key point at this analysis is to determine what caused human errors to be responsible for the failures both in time and hierarchy.
Especially Swiss cheese model has been proven to be very useful in order to avoid human errors. Professor J. Reasons concluded that human error stems from slips, lapses and trips. More specifically he has stated that reassembly of components is the biggest cause of problems. A frequent phenomenon is the omissions of some fasteners during the reassembly which can cause major problems. To tackle those omissions, 4 safety barriers were proposed which are:
Organisation, meaning that there will be some rules of the company that should be followed even if one thinks there is a better way
Supervision by assigning to another person, other than the responsible one for a specific task, to double check the numbers/operation. Penetration of this barrier can be realised by showing exaggerating trust on the responsible person from the side of the supervisor
Preconditions which is something on the background that you don’t understand
Active failure which is the time when you do the last error and the accident arrives. There are also latent failures which are the most dangerous.
If all these defences are breached, then a serious event is happening. A very recent risk situation is the pandemic of Covid-19. According to scientists, experts and policy makers, there are many barriers to limit the spreading as shown in the figure above created by Ian Mackay.
The first barriers refer to personal habits such as physical distance, masks, hand hygiene and avoiding congestion. The last barriers refer to shared responsibilities and include tracing of confirmed cases, ventilation of buildings, government policy and the last line of defence is the vaccines.
Coming back to wind turbine industry, it is imperative that human error should be limited during maintenance tasks. Wrong decisions have a huge negative effect. That’s why, automatic detection and identification of component failures using advanced processing techniques like Machine Learning and Deep Learning can reduce the effect of human errors and make maintenance a safer and more precise procedure.
Moreover, the reassembly which is a routine process for many components can be improved by introducing Augmented Reality (AR) equipment so as to keep technicians focused at what component follows after the placement of the previous one and limit the dependence of the technician on the manuals. Therefore slips, lapses mistakes and violations of rules and laws can be avoided using advanced technologies.
- Panagiotis Korkos, Doctoral Researcher